Qihoo 360 unveils Tulong Feng as China’s Mythos

Qihoo 360 founder Zhou Hongyi introduced Tulong Feng at ISC.AI 2026 in Beijing on June 24, calling it China’s answer to Anthropic’s Mythos. Separately, Beijing-based lab Z.ai released GLM-5.2 under an MIT license and made the model freely downloadable and modifiable.

Zhou presented Tulong Feng alongside an automated defense platform named Yitian Zhen and a new industry coalition called Panshi Zhidun, or Shield of Bedrock. He described Mythos as ‘cyber nuclear weapons’ and said Chinese firms are excluded from Anthropic’s Glasswing vetted partner program. Zhou reported Tulong Feng has flagged 3,432 cumulative vulnerabilities, with 105 confirmed by Chinese regulatory bodies and several entries marked high severity in the national vulnerability database. He said the system follows an agent-first approach that coordinates multiple specialized models rather than relying on a single large model.

Z.ai published GLM-5.2 shortly after U.S. authorities pulled Mythos 5 and Fable 5 offline for foreign users. The lab released the model under an MIT license, removing subscription and geographic limits and permitting public modification. Independent evaluations cited by Z.ai placed GLM-5.2 at a 39% F1 score on an insecure direct object reference detection test, ahead of a competing model on the same test, and found it matched a competing model on a capture-the-flag challenge. Z.ai estimated the cost per vulnerability finding at about $0.17, compared with more than $1 per finding using workflows based on the other model.

Z.ai co-founder Tang Jie called Anthropic’s withdrawal ‘deeply regrettable.’ Technical lead Qinkai Zheng added, ‘We want the model accessible to everyone.’ Tang rejected a public prediction that China would not reach comparable capability until early 2027, saying ‘Won’t take that long.’

The announcements follow U.S. export controls that restricted Anthropic’s cybersecurity models and placed access behind a vetted coalition. Anthropic is negotiating with the Commerce Department over reinstating access to Mythos 5 and Fable 5. Security officials and industry representatives have raised concerns about autonomous systems that can find vulnerabilities and assemble attack chains without human oversight. Developers at Qihoo 360 and Z.ai stated their tools are intended for defensive testing and domestic security work.